ED 349 957

AUTHOR_TITLE

INSTITUTIONSPONS AGENCY

PUB DATECONTRACTNOTEPUB TYPE

EDRS PRICEDESCRIPTORS

DOCUMENT RESUME

IR 015 693

Newman, Denis; And OthersLocal Infrastructures for School Networking: CurrentModels and Prospects. Technical Report No. 22.Center for Technology in Education, New York, NY.Department of Education, Washington, DC.; NationalScience Foundation, Washington, D.C.May 921-135562167-Al; MDR-915400630p.Information Analyses (070) Viewpoints(Opinion/Position Papers, Essays, etc.) (120)

MF01/PCO" Plus Postage.Communications; Comparative Analysis; *ComputerNetworks; Educational Change; Educational Technology;Elementary Secondary Education; *Futures (ofSociety); Information Networks; *Local Area Networks;Microcomputers; *Models; *Telecommunications

IDENTIFIERS *Wide Area Networks

ABSTRACTThis paper identifies a paradigm shift that must take

place in school networking. The ultimate goal is to retool theschools with a local technical infrastructure that gives teachers and

students immediate access to communication systems and information

resources, thereby supporting the implementation of advances in

pedagogy and educational technology. The current notion of

telecomputing cannot address the information requirements locallywithin the school and, ultimately, will fragment and inhibit any movetoward universal access to information resources. A technology is

needed that combines local and wide area networking (LAN and WAN),

making access to remote resources part of the everyday work withschool computers. This report contains the following sections: (1)

The Problem: Combining Local and Wide Area Communication--facts aboutthe current state of school networks and the dissociation of schoolLANs and WANs; (2) A Brief History of Network Technology; (3) A

Convergence of School LAN's and WAN's--integrating and simplifying aschool internetwork; (4) Current Models of School LAN-WANConnectivity--a comparison of six models; and (5) Prospects for theFuture. (Contains 10 notes and 34 references.) (Author/ALF)

*************************************************1c*********************

AeprucluLLIonS supplied oy Evic are tne best that can be made* from the original document.***********************************************************************

n

U.S. DEPARTMENT OF EDUCATIONOfirce of Educational Research and Improvement

EDUCATIONAL RESOURCES INFORMATIONCENTER (ERIC)

C This document has been reproduced asreceived from the person or organizationoriginating it.

C Minor changes have been made to Improvereproduction Quality

Points of view or opinions slated in this docu.merit do not necessarily represent officialOERI position or policy

CENTER FOR

(HNC 06!

n ras rut ures

Denis Newman,Suian

Bolt Beranek aticlNe#matt

Paul A. Reiti-Ralph Bundle Schad

BANK STREET COLLEGEOF EDUCATION

610 WEST 112th STREETNEW YORK, NY 10025

2

ET COPY AVAILABLE

"PERMISSION TO REPRODUCE THISMATERIAL HAS BEEN GRANTED BY

K. McHillan

TO THE EDUCATIONAL RESOURCESINFORMATION CENTER (ERIC)"

TechnicalReportSeries

Technical Report No. 22

:1041 Infrosirsktutis

for School Oit*ofking:,Orrin! Modelistindiispects

Denis New0m,, timess%

Bolt Birono4 and *wain ,,%.

Ralph Bunche School

May 1992

3

BEST COPY AVAILABLE

nR7eP LOCAL INFRASTRUCTURES

FOR SCHOOL NETWORKING:

Current Models and Prospects )

.

N>

Batt Bitiiiiek and Isievvrntin 'Ralph Bundle Sciti)Of

Abstract

This paper identifies a paradigm shift that must take place in school networking Our ultimategoal is to retool the schools with a local technical infrastructure that gives teaches and studentsimmediate access to communication systems and information resources and, thereby, supportsthe implementation of advances in pedagogy and educational technology. On a moreimmediate level, we want to revise an overly narrow conception that many educators holdabout network technology. In schools, the talk of "telecommunications" or "telecomputing"refers to the decades-old technology ofconnecting terminals to time-sharing computers. Ifthisconcept of networks persists, it will have a stultifying effect as schools attempt to use networksduring this decade. The current notion of telecomputing cannot address the informationrequirements locally within the school and, ultimately, will fragment and inhibit any movetoward universal access to information resources. A technology is needed that combines localand wide area networking, making access to remote resources part of the everyday work withschool computers.

Summary of Recommendations'Our analysis of the current state of school networkingleads us to specific recommendations for a program ofresearch and technology development. To take advan-tage of the existing and planned local infrastructures inschools and school districts and to assure the broadest

The research reported here was supported by the NationalScience Foundation under grant MDR-9154006. Additionalsupport for the preparation and dissemination of this reportcame from the Center for Technology in Education underGrant #I- 13 5562 167-A 1 from the U.S. Department ofEducation to Bank Street College of Education. The authorsare grateful to Margaret Honey, JoAnne KeiAen, Cecelia Lenk,andJohn Richards helpful comments on earlier drafts ofthispaper.

possible implementation, four areas of technical de-velopment will be necessary:

As an alternative to the pervasive terminal cos -nections in school telecomputing, remote net-work connections (direct connections betweenschool computers and a national educationalnetwork) must be supported and made easy-to-use.

Before using the extensive networks already inplace for school administrative functions, schoolnetwork technologies designed for instructionand teacher support must address network secu-rity and management.

To assure that network services are both afford-able and appropriately tailored to schools, a

4

Tech. Rep. No. 22 May'1992

network-server technology that can be distrib-uted within schools, school districts, and re-gional networks must be developed.

Within such a network of distributed servers,there is a need for special-purpose "client" soft-ware that simplifies students' and teachers' inter-

actions with network services.

These developments must occur in the context ofa program of research that will allow us to refine thedesign and application of advanced network technol-ogy for schools, which have their own requirementsand organization.

Contents

The Problem: Combining Local and Wide Area Communication 3Why Networking? 3Why This Project? 3Some Facts About the Current State of School Networks 3The Dissociation of School LANs and WANs 5Problems to Address 5

A Brief History of Network Technology 6Centralized Computing 6Distribrtted Computing 7Value of" Distributed Computing 8Local Area Networks 8internetworking 9Client-Server Systems 10Modes of Access to the Internet 10

A Convergence of School LANs and WANs 11The Elements: LANs, WANs and "DANs" 11A Scenario 13Integrating and Simplifying a School Internetwork 14

Current Models of School LAN-WAN Connectivity 15A Matrix of LAN-WAN Connections 15Comparing the Six Models 20

Prospects for the Future 23Build Remote Network Connections 23Address Security and Management Concerns 23Develop an Integrated Server 24Develop Client Software 24

Notes 25

References 26

2

May 1992Tech. Rep. No. 22

The Problem: Combining Local and Wide Area Communication

The research we report in this paper explored methods for using communication technology

within the school to augment access to remote information resources. The problems we sought

to address were in some ways more deeply rooted than we expected. Yet the opportunitiesfbrimproving education remain a powerful incentivefor developing a new paradigm for school

networking.

Why Networking?In spite of widespread publicity for telecomputing,communication technology of any sort is almost en-tirely absent from our schools. Even schools that have

invested heavily in computers have surprisingly littlein the way of telephones or computer networks avail-able for instructional uses. Traditional educators mayask: Why should they?'A set of textbooks for a class-

room can effectively encode the required instructionalcontent. Computer technologies can deliver effectiveexercises. Interactive multimedia can even engagestudents in motivating stories and visual presenta-tions. What can communication technology add tothis arsenal of instructional delivery methods?

The answer to the instructional delivery paradigmcan be stated quite simply. Information required forproductive citizenship is changing rapidly. It is nowessential to present instructional content that is live,

that comes from active information sources such asweather satellites. data collected by students in otherlocations, and responses from working scientists. Schoolwork must include primary information sources sothat students are simultaneously learning the critical

content while gaining firsthand experience with infor-

mation sources themselves.At the same time, information access technology

is evolving rapidly and dropping in cost. Changes inthe teachers' role from a dispenser of information to acoach who helps guide students' firsthand experiencewith the live sources of information will require sup-port of teachers in finding and using these sources; ingetting current information about technologies that

may be helpful to them in the classroom; and indeveloping techniques for using these technologies in

the classroom.

Why This Project?The potential of communications technologies for

opening schooling to information sources is recog-nized as one of the "Grand Challenges" of the HighPerformance Computing Act of 1991 (Office of Sci-

ence and Technology Policy, 1991). In coordinationwith the National Research and Education Network(NREN) program,' the National Science Foundation(NSF) has initiated a program to develop the ed-ucational potential of computer communication net-works (Hunter, 1992a). At the same time, manystates, including Texas, Kentucky, New York, andCalifornia, have begun the development of statewideeducational networks.

The NR EN is expected to be shared not only bythe higher education and scientific and research com-munities, but also by K-12 and related educationalinstitutions and, indeed, by the larger public (e.g., seeMelmed & Fisher, 1991). Given the national im-portance and opportunity of the NREN, we assume inthis report that school networking and the technologi-cal infrastructure on which it is based must allowparticipation in the future NREN (see Hunter, 1992b).

In the context of this identified opportunity toimprove schools and these converging federal andstate programs, we obtained funding from NSF toundertake an informal study of the local infrastruc-tures in the schools and school districts that mightsupport widespread, perhaps universal, access to anational school network as an extension of the NREN.

Some Facts About the CurrentState of School NetworksWe were well aware in starting this work that mostschools that use a telecommunications or wide areanetwork (WAN) have a single computer with a mo-dem attached to a phone line, perhaps in one of theclassrooms or at the teacher's home.

It seemed obvious that this single resource wouldnot be adequate for thewhole school. At best, it could

serve a handful of classes. Most teachers and studentswould have no firsthand experience with informationsources outside the school.

We were also aware of a handful of exampleswhere the school used a local area network (LAN),connecting either all the computers in a computer lab

6 3

School

E Modem

iketele

or connecting computers 'distributed among class-rooms, for communication within the school. In somerare cases, the local network was also used to broadenaccess to outside resources, providinga network connec-tion for every classroom computer. This is the equiva-lent of every computer having its own modem andphone line. These LANs give many more students andteachers access to the WAN than is possible with justa single point of connection (and do so far moreefficiently than would be possible by giving everyclassroom its own phone line and modem).

Our plan w.s to document such cases to providemodels that other projects could emulate. A survey ofCalifornia schools, conducted by the California Tech-nology Project (1990), provided us with a glimpse ofthe current state of school networks. Their surveytabulated responses from 485 schools.' It includedquestions as to whether the school had a LAN andwhether there was a modem or a wide area network

May 1992

connection available. For purposes of our report, weobtained an additional analysis of the data from CTPto find the intersection of responses on these ques-tions, allowing us to construct the following matrix forthe 407 schools for which appropriate data was avail-able:

YES

NO

Wide Area NetworkConnection

YES NO

43 24(10.6%) (5.9%)

140 200(34.4%) (49.1%)

While 16.5% of the schools have LANs and 45%have a wide area connection, only 10.6% have both.This 10.6% is a good place to start, since here we mayfind the schools most advanced in building on their

resources to make best use of their wide areaconnection. We were able to contact 30 of the 43schools in this set. The striking outcome of this follow-up research was that none of these schools made use oftheir instructional LAN for distributing the data com-ing from the WAN.3 In more than half the cases, themodem used in instruction was not even on a LANcomputer.

The picture is a bit different if we look at admin-istrative applications, such as reporting attendancerecords or administrative communication in the 30schools. Twenty-one of the schools had a modem foradministrative or management functions and 12 ofthose were connected to a LAN. In seven cases, theschool administration office had a LAN of computersthat served as terminals on a district database. In fivecases, the modem was on an instructional LAN, butwas used as a channel for the vendor to download newsoftware versions and to perform troubleshooting.Clearly, the technology for connectingLANs to WANsexists in the schools, but it is not being applied outside

4 7

of administrative functions. It is not available toteachers and students.

The Dissociation of School LANs and WANsThe almost complete absence of connections betweeninstructional LANs and instructional uses of WANconnections is indicative of an ingrained conceptualdissociation between the two technologies.

On the one hand, LANs are used predominantlyas a way to distribute software, as a means for sharingprinters, or as a mechanism for managing instructionand saving records of student work.

On the other hand, telecommunications or"telecomputing" networks are a conceptually unre-lated technology in which phone lines are used toconnect a school computer to a community or infor-mation source outside the school.

Paradoxically, in business, research, and highereducation, the primary use of LANs is for communi-cation and sharing data. There, the technology forconnecting LANs to WANs is well established. It isnot easy, however, for a school to simply purchasesystems designed for the business setting. Comparedto office environments, schools are very large, com-plex, and volatile, with students changing grades andclasses every year. While this puts a greater burden onsystem administration (e.g., creating and modifyingstudent mailboxes, work group assignments), schoolshave fewer resources for full-time system operatorsthan do other institutions. A network system forschools will require very high ease-of-use and a man-agement system that integrates network services andstudent records.

The lack of LAN-WAN connections for teachingand learning in schools is not due merely to a lack ofresources. Many LANs are in place. There are schoolswith wiring in place for their intercom systems thatcould serve adequately for a reasonably sophisticatedLAN connecting all classrooms. Many school districtshave WANs in place for administrative functions.However, the distinct functions of school LANs (asinstructional management and delivery) and WANs(as communication and resource access) has resultedin a technological gap. Currently, none of the localarea network vendors selling specifically to schoolsoffer a product that is appropriate for connecting aninstructional LAN to a wide area network. Vendors ofadministrative systems understand the connectionbetween the local and the wide area networks but the

concept has not been applied to instructional uses.

Problems to AddressDeveloping communications functions within schoolsis necessary if we are to avoid the following conse-quences of the current paradigm:

Lack of access. Without any means of transmit-ting data, electronic mail, and documents aroundthe school, the constraints of scheduling accessto a single network terminal will limit the num-ber of network projects that a school can engagein and the number of students and teachers whocan have significant contact with the networkresources.

Inequities. Existing school infrastructures willfavor thc students who are already privileged.The limited access to network projects will likelyfavor students who are assigned to enrichmentactivities. In addition, inner-city schools aremore likely to acquire local network systemsproviding basic skills lessons rather than com-munications or data sharing capabilities (Cole &Griffin, 1987; Sherry, 1990).Marginal impact on the school. A local infrastruc-ture can multiply the impact of the wide areanetwork on the school. The local intellectualcommunity is the foundation on which to buildactivities that benefit from the network resources.

A serious development effort will be needed to fillthe technological gap. More important, perhaps, aresearch and demonstration effort will be needed toarticulate an alternative model. Few school systems arelikely to take advantage of their installed local infra-structure for instruction or staff development unlessthere is a compelling pedagogical rationale. Givingteachers and students widespread, equitable access tothe NREN resources could be a rationale for investingin the small additional costs of providing flexible andequitable access to all students and teachers.

In the next section of this paper, we present a briefhistory of network technology to locate the origin ofthe unusual dissociation of local and wide area net-works in schools. We then outline some of the modesthat we have found for schools to connect their localand wide area networks, with special attention to howschools may eventually move towards providing broadaccess to the NREN. Finally, we argue the case foreliminating the more commonly found telecomputing

5

connection between school modems and outside re-sources in favor of a true network connection of thesort that is used in industry and in many school

administrative networksone. that will provide themost appropriate access to the NREN.

A Brief History of Network Technology

The dissociation of local and wide area networks has its roots in the historical introductionof the technologies in the schools. It might help to back up and take a look at the history ofcomputer networks. This section serves as a tutorial on the technology that will help frame oursubsequent discussion of school networks. Perhaps we can reduce some of the confusion ofterminology that has made it difficult to understand concepts like "the NREN" and hasgottenin the way of seeing the commonalties of local and wide area technology.

When educators talk about a "computernetwork," they are most often referringto a network service. This usage is re-flected in several current surveys of

school and educational networks (Kurshan, 1990a,1990b, 1991; McAnge et al., 1990). For example,many teachers subscribe to CompuServe, AmericaOn-line, Prodigy, or AppleLink, which are informationservices offered via a dial-up connection. Similarly,classrooms subscribe to the National Geographic KidsNetwork or to the AT&T Learning Network, whichare more structured products provided by similarservices. Local or dial-up bulletin board systems (BBSs)are also sometimes called networks. At other times,"computer network" refers to a community of peoplewho use access to a particular service as a means ofcommunication. For example, the "superintendents'network" consists of a group of administrators whodial into a computer at the Merrimack EducationCenter in Massachusetts to communicate by elec-tronic mail and share database resources. This com-mon usage implicitly accepts an historically anti-quated notion ofcomputer communication because itassumes that there is a particular computer service orhost computer that all the members of the communityuse. Their common connection to that host is whatdefines the community. To understand the history ofnetworks, we have to get below the level of the servicesand communities that make use of networks and lookat the networks as systems of hardware, software, andtransmission media.

Centralized ComputingThe first thing resembling a computer network wasthe terminal-to-host system used on timesharingcorn-

puters. First developed in the sixties, the timesharingcomputer (which became known as a "host") allowedmany people to access it simultaneously from termi-nals (desktop consoles) wired directly to the com-puter. In the late sixties and early seventies, the avail-ability of modems, which turn digital into analogsignals and vice versa, allowed the terminals to belocated at the far end of ordinary voice phone lines.Those phone lines could be dedicated (i.e., leased fromthe phone company for exclusive use by the networkowner) or switched (i.e., dial-up use of the publicphone system). The host always managed the termi-nals and performed all computing.

The model of centralized computing and dedi-cated terminal-to-host communications can be repre-sented as a simple star-like figure.

When personal computers (PCs) became wide-spread a decade ago, "terminal emulation" programswere developed so that a PC could serve as a terminal.This paradigm was elaborated with protocols fur send-ing files back and forth and with "user interfacesoftware" for PCs that let the PC present the menuoptions using its own graphical interface. But theessential feature of this paradigm is that the PC isconnected as a terminal to one specific host.

6

May 1992 Tech. Rep. No. 22

Distributed ComputingWhile the terminal-host connection let remote usersuse a single host, it did little for hosts that wanted tocommunicate with other hosts. Packet switching wasdeveloped to solve the problem of inter-host commu-nication. If a database were developed for one host,how could it be transferred CO another host? Howcould a database administrator sitting at a terminal onone host get files from or send files to another host?How could a scientist working on one host run aprogram that was on a host at a distant university? Thedevelopment of packet switching technologywas givena big boost in the late sixties and seventies with theU.S. Defense Department-sponsored ARPANET, thepredecessor of the NSFNET (Corner, 1990; Dern,1989) and what is now commonly called the Internet.Packet switching was developed as an efficient way formany users to access and transfer data among manytimesharing computers from different manufacturers,and for those computers to share expensive resources,that is, the long distance telephone lines' connectingthem. ARPANET hosts were not physically con-nected directly to each other, but were connected tospecial-purpose computers called packet switches. Wecan represent this distributed computing and com-puter-to-computer communications as a complex meshnetwork with packet switches (squares) serving to getthe packets from one host to another (dark circles).Conceptually, the hosts form a network of peers, eachon an equal footing.

Termirs.,

Packet switches--

The ARPANET demonstrated how computingresources could be widely distributed on a reliable,shared, wide area infrastructure.5 Each computer wasa peer of the others; each had a network address andcould send and receive messages.

Packets

The basis of packet switching is the concept of a packet.In the usual terminal-host connection, a stream of bitsis transmitted between the terminal and host. Sincethe line between the terminal and host is dedicated tothat connection (in the case of a dial-up, it is tempo-rarily dedicated), there is no question as to where thebits are supposed to go. Anything from the terminalgoes to the host and vice versa. In packet-switchednetworks, the bits from any one computer may wantto go to any of the other computers, so there had to bea way to address the bits appropriately. Assembling apacket is simply like putting a group of bits in anenvelope and addressing it to another computer. Eachhost has an address on the network. As the packets ofdata are sent out over the network, packet switchesdetermine the most efficient route to the other com-puter.

In addition to the seminal ARPANET technol-ogy, several approaches to packet-oriented network-ing have emerged over the last decade and a half. Forexample:

DECNet and SNA are the proprietary networkarchitectures of Digital Equipment Corpora-tion and IBM, respectively.

X.25 is the international standard computerinterface to packet switching networks and isused in all public data networks such as Telenetand Tymnet, as well as by private companies andorganizations.

OSI (Open Systems Interconnection) representsa set of international standards for inter-connecting different manufacturers' systemsworldwide (X.25 is part of the OSI scheme).

TCP/IP6 emerged out of the work on theARPANET and is now the basis for the Internet(see discussion of internetworking below).

TCP/IP, X.25 and OSI are independent of spe-cific computer manufacturers and their products, incontrast to SNA and DECNet. Irrespective of thesedistinctions, this heterogeneous collection of methods

10 7

ec . ep. 'a. May 1992

_ for addressing packets represents the real beginning ofcomputer networks.

Store-and-forward networks

It is useful to distinguish a class of distributed net-works commonly used in school telecommunicationsbut not based on packet-switching technology. In-stead, these networks use communications protocolsfor message transfer which are designed for "batch" filetransfers rather than interactive traffic.

FIDOnet, including the subset called K 1 2ner,consists of thousands of home and school micro-computers running electronic Bulletin BoardSystem (BBS) software; the BBS messages andfiles are distributed ("echoed") from one toanother micro (FIDOnet BBS) over dial-upphone lines when rates are lowest (generally atnight); each BBS subscribes to a set of echoes ortopics (Reilly, 1992).

FrEdMail (Free Educational Electronic Mail) issimilar in structure to FIDOnet, consisting ofabout 125 computers that exchange messagesover dial-up lines on a scheduled basis. Thecomputers are run by volunteers and offer dis-cussions and collaborative projects for teachersand students.

BITNET ("Because It's Time Network") is acollege and university network of host comput-ers (BITNET nodes) that transmit messages andfiles between each other using IBM protocols for"batch" submission of computer jobs (LaQuey,1990).

These networks, generally called "store-and-for-ward," distribute the computing, but communica-tions with end users arc not interactive or "real-time"and are limited to the exchange of text files.

Value of Distributed ComputingThe centralized computer approach could never haveachieved what has happened with distributed, packetswitching networks. A single, centralized computercould not have held all the computing and dataresources of the ARPANET hosts. And connections tosuch a computer from all the people wishing to use theARPANET would have been unmanageable. Bydistributing computing around the country and evenaround the world, the system as a whole is more

reliable and robust, less expensive per user, easier toadd users to, and can support many more applicationsand databases.

Packet switching made possible the developmentand subsequent astounding growth of electronic mail("email"). The ability for people to communicate aspeers using electronic messages was built on top of theanalogous capabilities in the distributed computernetwork. Wide area communications began to signifyhuman interchange, not merely computer data ex-change.

Local Area NetworksIn the early stages of the ARPANET, people were stillusing terminals to connect to their research center'shost computer via a phone call or direct wire. The nextmajor advance in computer networks occurred withlocal area networks connecting PCs within a buildingto printers and other resources, including host com-puters. By the early 1980s, local area networks (LANs)were becoming more commonplace on campuses andin corporations, as losi/-cost desktop systems (PCs)and sophisticated desktop workstations became avail-able.

PersonalComput rs

Host

Packet switches-4.-

A large number of different LAN technologiesnow exist; for example, AppleTalk and Novell(NetWare) are widely used. Lower level LAN trans-mission schemes include the standardized Token Ringand Ethernet, and Apple's LocalTalk. Special purposehardware (generally called a gateway) is used to con-

81.1

'PACly 1992 Tech. Rep. No. 22

nett different physical LANs that may be found in thesame building; for example, LocalTalk and Ethernet.LANs are commonly wired out of coaxial cable orplain telephone wire or, more recently, optical fiber,

Like the wide area distributed networks, LANsalso used addressed packets to transmit informationfrom one PC to another. A local host could also besituated on a LAN. Now the connection between thedesktop PC and the host was just part of the localpacket network.

Internetworking'The development of different wide area network tech-nologies (such as packet radio) and the need to inter-connect multiple packet networks each using differentprotocols led to the growth of internetworking. Inter-nerwork technology allows multiple diverse comput-ers on different physical networks to communicate,using a standard set of communications conventionsor protocols.

The internetworking research, sponsored by theU.S. government in the early 1980s, was a directoutgrowth of ARPANET technology. Internetworkpackets were routed to their destinations by special-purpose computers called routers, which are a form ofpacket switch and serve similar functions in a networkof networks. Using Internet protocols, workstations(and, more recently, PCs) on LANs could commu-nicate in exactly the same manner with each other, andwith distant computers via wide area networks as theydid with other computers on their own LANs.

Within the internetwork, all the computers(whether PCs or mainframe hosts) were equal in thesense that any computer could communicate with(i.e., address packets to) any other computer. Now,not only were more remote resources available morewidely, the ability to make use of those resourceslocally (i.e., on the user's desktop computer) increaseddramatically.

In addition to electronic mail, the most widelyused network services that flourished with the devel-opment of distributed networks included remote loginusing terminal emulation, file transfer, and networknews. The TCP/IP protocol suite includes protocolsthatsupport these services. SMTP (Simple Mail Trans-fer Protocol), Telnet (for remote login), FTP (FileTransfer Protocol), and NNTP (Network News Trans-fer Protocol) are standard protocols. Telnet and FTPare also the names of application programs that users

invoke, and have assumed the status of verbs (e.g., "totelnet to host A"; "to ftp the file"). Most Internetlocations implement the distributed bulletin boardsystem known as the USENET or netnews, and sub-scribe to newsgroups.' Research in internetworkingand its widespread dissemination via distributed net-works have led to many more network-based servicesin recent years.

What is now called "the Internet" (with a capitalI) is an important development because it has ex-tended this notion of internetworking internationally.The Internet consists of the global set of intercon-nected TCP/IP networks that share a common ad-dress space. Historically, the Internet has also beenknown as the ARPA or DARPA Internet (ARPANETwas the first network in the Internet) and the TCP/IPInternet (see also Cerf, 1989). Today it consists ofmore than 5,000 networks, which link together hun-dreds of thousands of computers and millions of usersthroughout the world. The domestic, nonmilitaryportion of the Internet includes NSFNET, whichserves the U.S. scientific and engineering researchcommunity. NSFNET has a 3-tier structure com-posed of

a "backbone" connecting the NSF-fundedsupercomputer centers, It consists of a series ofpacket switches (routers) and very high-speedlink!,. Like an interstate highway, the backboneaccepts traffic from any of the regional networksor supercomputer sites;

mid-level or regional networks, such as T'HEnet(Texas Higher Education Network), NYSERNet(New York State Education and Research Net-work), NEARnet (New England Academic andResearch Network), and CERFnet (CaliforniaEducation and Research Federation Network),which are autonomously administered;

campus and corporate networks which connectto ("are members or) the regionals and whichcan be quite extensive themselves.

The Internet also includes statewide networkssuch as CSUnet (California State University's net-work) and other federally sponsored networks such asNASA Science Internet (NSI) and Energy SciencesNetwork (ESnet), and connects TCP/IP networks inmany countries and regions of the world (see LaQuey,1990). The high-speed National Research and Educa-tion Network (NREN) is projected to evolve from a

12 9

part of the Internet containing portions of the currentNSFNET, NSI, and ESnet.

Client-Server SystemsWhile all computers on a distributed network areequal in one sense, the notion that some computerscould or should provide services to other computersemerged. The "client-server" technology combinessome of the virtues of centralized computing' with thestrengths of distributed networking (Wood & Mensch,1991). Certain computers are dedicated as mailboxes(mail server) or for file storage (file server) or tomaintain databases (database servers). Other com-puters, usually the PCs on people's desktops, are the"clients" in this paradigm.

Combined with internetworking, these client-server systems are now becoming commonplace inbusiness and research settings. The desktop PC can bepart of a world-wide Internet with servers in the sameroom or at a remote university. Routers on the LANtake care of sending packets out over the wide areaconnections. More significantly, in client-server com-puting, the person at the desktop PC is not personallyinteracting with the server. It is the client softwareembedded in the person's PC application that requestsservice from server software. The desktop PC is nolonger doing terminal emulation. To the user, thecommunications involved in accessing mailboxes, files,or databases are invisible and nonintrusive. It is "as if"these mailboxes, files, or databases were on the user'sdesktop machine. The network pervades the person'sinteractions with the PC.

The convenience of client-server systems has re-cently appeared in the context ofcentralized terminal-host connections. CompuServe, for example, has aprogram called Information Manager that providesicons for each of the services. Rather than selectingfrom a menu presented by the host, the subscriber canclick on an icon on the PC "desktop." Similarly, theNational Geographic Kids Network provides a clientprogram that makes the call, and sends and receivesdata without the teacher's having to select items froma host-based menu. The same program allows thestudents to display and analyze the data.

While making the work of interacting with thehost much easier, these so-called "user-interface soft-ware" packages retain the weaknesses of the terminal-host paradigm. The fact that these programs are notbased on packet - network technology may make little

difference as long as the connection is restricted tothose two computers. The centralized computingparadigm underlying the connection, however, makesit difficult to integrate a local area network or toflexibly address resources distributed on a wide areanetwork. Client-server systems based on packet-net-work technology, on the other hand, have great prom-ise for education since they simplify interactions withremote or local servers while retaining the flexibilityand scaleability of internetworking.

Modes of Access to the InternetA local area network is not required in order to attacha classroom computer to a wide area packet-switchednetwork. For example, a standard called Serial LineInternet Protocol (SLIP)9 makes it possible to connecta single computer with a modem and ordinary phoneline directly to the Internet. These connections re-portedly require some of the patience traditionallyassociated with telecommunications and are not there-fore appropriate for unsupported, novice networkers.If school PCs are on a LAN, the work of connectingthe LAN and its classroom computers to the Internetis done either by a dedicated router or by routingsoftware on the local server.

Schools lag behind universities and industry intheir use ofInternet connectivity options. In almost allcases where teachers have "access to the Internet," theyare still using a traditional terminal-host connection.Just as in universities and in industry before theinstallation of LANs, a terminal connection is made toa host located at a local university or other Internetmember or maintained by a state education networksuch as Texas Education Network (TENET).

This host may provide bulletin boards, electronicmail, databases, and many other services to peoplewith user, accounts. Since this host is part of theInternet, the teacher may have permission to runprograms on it to remotely log into any other com-puter on the Internet, transfer files to and from otherInternet computers, send and receive mail, and so on.But it is the host computer that is performing thesefunctions, not the computer on the teacher's desk.The teacher's PC is still just a terminal on the host.

So, for example, the teacher might request that agraphic file from the Magellan mission be transferredfrom the NASA computer (on the Internet) to thelocal host he is logged into. A few seconds later, the fileis on the local host. Now, he must use the terminal

10 13

. e .

software of his PC to "download" the file across thephone line. This two-step process contrasts with thescientist whose PC is on a LAN that is part of theInternet. In that case, she requests the NASA com-puter to transfer the file, and a few seconds later the fileis on the disk of her PC.

Our simplified history of network technology hascharted the substantial progress that has been madesince the days of centralized, proprietary computingsystems. Network technology is increasingly distrib-uted, integrated, and based on open public standards.Local and wide area networks have become extensionsof one another.

A Convergence of School LANs and WANs

School networks are now at an interestingjuncture. Only a handful are doingtrue networkingin the sense of having their school computers connected to a wide area packet-switchednetwork. But school LANs are being used increasingly for communicating and sharing datalocally, and the interest in the Internet as a vehicle forschool nettvorking suggests the potential

for extending internetworking and client-server systems down to the student and teacherdesktop computers.

Sharing andCommunication

InstructionalDelivery

NetworkConnections

Terminal-hostConnections

The convergence of LAN and WAN systemsis at a very early stage. But we see criticalelements in place that will support a newparadigm. In this section we describe the

basis for the convergence and illustrate it with ahypothetical scenario.

The Elements: LANs, WANs and "DANs"Local area networksThe original educational experiments with timeshar-ing computers, such as the Plato system, were orientedto instructional delivery. This orientation has carriedover to the so-called Integrated Learning Systems(ILSs) (see Sherry, 1990). Technically, many of theILSs are LAN-based client-server systems and their

evolution from terminal-host systems parallels theevolution in the rest of the computer industry.

Generally school LANs are seen as managementtools that serve the purposes of teachers in charge ofoperating a computer lab, running a remedial pro-gram, and soon. Management systems, such as ICLASfor IBM PCs running on a Novell LAN or Aristotle forApple Its on an AppleTalk LAN, provide rudimentaryclient programs which simply display a menu of thesoftware that the student at the computer is allowed touse in that session in the computer lab. This lets the

14 11

Tech. Rep. No. 22O

May 1992

teacher in charge easily control software usage whileavoiding the annoyance of floppy disks.

While still few in number, some schools are begin-ning to use LANs to support tool-based student projectswhere the file server is used as a storage medium forproject-related files and for communication amongteachers (and occasionally among students). For ex-ample, in the Earth Lab project at New York City'sRalph Bunche School, students make extensive use ofword processing, databases and electronic mail(Goldman & Newman, in press; Newman, in press;Newman, 1990; Newman, Goldman, Brienne, Jack-son, & Magzamen, 1989; Newman & Reese, 1990).Each student, as well as each work group, has anelectronic laboratory for the project - related text, data,and other files. The laboratories are folders (directo-ries) on the file server that serve as group or individualworkspaces.

The newspaper editorial group, for example, has alaboratory in which issues of the newspaper are as-sembled from stories and pictures contributed bystudents from various classrooms. Electronic mail isused among students for both personal and school-related projects. In this case, the LAN file server is nolonger a centralized point of control but a mediator ofcommunication among students and teachers. Theuse of school LANs for communication and informa-tion sharing may increase with the distribution ofcomputers around the school rather than in computerlabs, a direction in which some ILS vendors are

moving (Mageau, 1990; Electronic Learning, 1992).As individual teachers tailor computer use to their owninstructional units, taking advantage of continuousrather than episodic access to the computers, central-ized management will become less important. Withcomputers distributed around the school, the com-munication function of the LAN will become moresalient and the potential for using the LAN as amechanism for access to resources outside the class-room and outside the school will become more obvi-ous.

Wide area networksSince the late seventies, host computers have providedservices to schools, classrooms, and communities ofteachers (Kurshan, 1990a; Roberts, Blakeslee, Brown,& Lenk, 1990). The teacher or the school subscribesto the service, which may include conferencing withother teachers, weather reports, electronic mail stu-dent pen-pal communication, bibliographic data re-trieval, and so on. Consumer services, such asCompuServe, America On-line, or Prodigy, charge aconnect-time fee. Bulletin boards may offer similarservices on a smaller scale and operate free of chargeexcept for the phone call. All these services make useof a terminal-host connection with the classroomcomputer (or teacher's home computer))° Currently,it remains the case that practically all telecomputingprojects use this kind of connection.

The recent interest in using the Internet as avehicle for K-12 education is inviting a change in thetelecomputing paradigm. Sch(,3Is are beginning totake advantage of the distributed n2 ture of the net-work to access data and services around the world. Afew schools are now beginning to take advantage of thepacket-switched natuie of the Internet to make net-work conn,ections between their PCs and the network.As LANs become increasingly used for communica-tion and as the WANs of interest to schools aredistributed packet-switched networks, the two tech-nologies begin to converge, as we detail in the nextsection of this paper.

"District area networks"The vertical integration of LANs and WANs alsoinvites us to consider district networks, which cur-rently are used extensively for administration (e.g.,transferring attendance records to the district office).A district network, or what we might call a District

1215

Area Network (DAN), can serve as intermediary be-tween the school LANs and a state or national educa-tion network, such that connections between schooland national communities piggyback on the con-nections required in any case for administration. Thedevelopment of in ternetworki ng techniques, includ-ing the integration of IBM SNA technology, which isused extensively for district administration networkswith Internet technology (Guruge, 1992), lets usavoid costly duplication ofadministrative and instruc-tional networks. With a packet-switched network,functions at different levels of educational organiza-tion (school, school district, county, state, and na-tional) can be integrated into the same network.School LANs are an extension of this internetwork,reaching the teachers and students in the classrooms.

A ScenarioHere is a concrete picture of how this vertical integra-tion of a LAN, DAN and WAN might work. We startin a hypothetical classroom with an individual studentand a group of students working on their scienceinvestigations.

One student is making use of data stored on theschool server, while a collaborating pair ofstudents hasperformed an experiment on the velocity of a satellitein orbit and is about to send the results to a sister schoolin another state. The simulation program they areusing sends the results, along with their text annota-tion, as mail to the other school's mail server via a seriesof routers at various levels in the system.

LANconnectingall schoolcomputers _

The next picture illustrates schematically theschool's LAN, which connects all their desktop com-puters and servers. The server handles the students'communication locally as well as with schools aroundthe world, serving as an Internet router as well as a fileand database server. It stores the students' project data.

School

ServerFilesMailDatabaseInternet router

Remotenetworkserver

DistrictAreaNetwork

It also supports a management system that allowsthe school's computer coordinator to maintain theseservices for students and teachers in the school and toset privileges for network use, such as allowing thesestudents to transmit experimental results to the sister

411.11P'

BEST COPY AVAILABLE 1613

e441.1ep. No. 22

school and to access relevant Internet databases.The school network has additional specialized

modems serving as remote network servers, allowingteachers, students, or parents to communicate withthe school. For example, the computer coordinator isable to manage the server remotely from home.

The connection between the school and the dis-trict office runs over an ordinary telephone line form-ing part of the district area network. The next pictureshows that the same basic structure is replicated at theschool district office. The district server continues toroute the students' email to the state education net-work. The routing service at this level connects schoolswithin a district to each other, as well as through thestate educational network to other schools.

School

DistrictAreaNetwork

School

School

District office

ServerFilesMailDatabasentemet router

High speed=dam

School

ElMode

emotenetworkSWIM'

StateEductionalNetwork

The district office also has a LAN providingcommunication within the office. The district serverprovides the same services as those found on the schoolserver except here it contains aggregated data onstudents and other district administrative informa-tion. Not all the schools in the district have internalLANs yet. However, with an Internet router at thedistrict office, they are still able to tap into the Internetfrom at least one computer.

The connection between the districts and the stateeducation network center is shown in the next figure.The students' email message containing the results oftheir velocity experiment is one of very many kinds ofcommunication going back and forth across that line.At this level the district has leased a line to obtain a

May 1992

continuous high speed connection that is sufficient tohandle the level of communication being generated byall the schools.

The hierarchical nature of the Internet connectiv-ity scenario depicted here provides a smooth path forupgrades in access and capacity. School districts arenow in the situation that many colleges were in severalyears ago when Internet access was achieved throughintermittent dial-up lines. A tremendous amount ofinformation can be sent over an ordinary dial-upphone line." As the schools begin making heavier useof multimedia documents or synchronous communi-cation, capacity will have to be increased to avoiddelays.

Districtoffice

Districtoffice

Districtoffice

Districtoffice

/11111111

StateEducationNetworkCenter

NationalInternet

Integrating and Simplifyinga School InternetworkA major advantage of the distributed computing para-digm and ,the mechanisms for internetworking thathave grown up around it is the capability of bringingmultiple diverse networks into a single system. It mayseem paradoxical, b, t in fact the centralized comput-ing paradigm tends toward fragmentation. Each hostis associated with a separate community, has its ownway of interacting with its members, and often usesproprietary software. Internetworking, on the otherhand, is built on explicitly open (nonproprietary)mechanisms in order to be able to handle the diversityof computer platforms that want to be connected.

Compared to a proprietary network service orwell-run bulletin board, the Internet appears to many

14 1 7

teachers as chaotic and intimidating. Some attemptshave been made to make the resources more accessiblethrough handbooks such as The NYSERNet NewUser's Guide (1991) (see also Kehoe, 1992; Martin,1991), but even these require a fair level of sophistica-tion. The comparison of the Internet to a networkservice is inappropriate because the current difficultiespresented to the novice are a direct result of thedistributed nature of the network that makes an ever-expanding number of resources available.

Ultimately, bringing all educational resources intoa distributed network will tremendously increase thevalue of a connection to the network. But it will benecessary to develop for schools a client-server ar-chitecture in which the software running on the PC,not the person, knows how to interact with the remoteresources. The problem will not be to create central-

ized services, but to continue the migration of com-puting from the host out to the PC.

The convergence of school LANs used for sharinginformation and packet-switched wide area distrib-uted networks is opening up new possibilities that webelieve are worth exploring. As long as educators viewthe Internet as just another (unfriendly) networkservice or (incoherent) network community, we willnot succeed in tapping its potential, and the ways inwhich it invites connections to school LANs as part ofan internetwork will not be fully appreciated.

In the next section, we describe current models forconnecting LANs to wide area networks (not just tothe Internet). The experiments now getting under wayin schools can help to determine paths that schools cantake from the current paradigms to full participationin the national network.

Current Models of School LAN-WAN Connectivity

Over the lastyear, ourproject conducted an informal international search forschools that hadany kind of connection between their instructional LAN and instructional uses of wide areanetworks. In this section, we analyze the cases we found in terms of a matrix of six distinctmodels.

Recall that among the more than 400 schoolsin the California Technology Project survey,we were unable to find any instances ofconnections between instructional LANs and

instructional uses of wide area networks. We sentmessages out to a variety of network communities onthe Internet, AppleLink, and FrEdMail. The Kidsnetmailing list on the Internet yielded a good set ofcandidates for follow-up interviews. We also asked allthe major school LAN vendors12 for candidates (yield-ing no results). And we made inquiries of several of theInternet regionals with somewhat better luck. In all,we talked to people at approximately 40 schools. Wecertainly did not talk to all schools with LAN-WANconnections, but our impression was that we hadcontacted a substantial portion of the schools withsuch connections. Our impression is also that thenumber is growing rapidly as many schools and schoolsystems are making plans for Internet connections.Many schools we talked to considered their currentset-up an interim solution and had plans for, ordreams of, upgrades. LAN-WAN connectivity inschools is clearly in its infancy.

The following discussion is not meant as a how-tomanual nor will we address cost, an area of rapidchange. Groups currently addressing these issues in-clude the Internet Engineering Task Force workinggroup on K-12 networks and the Consortium forSchool Networking (CoSN) (see, e.g., Solomon, 1992;St. George, 1992). Our goal is to determine paths forgrowth for schools and school systems as they movetoward connections to the Internet. We expect to raisemore questions than we answer, but we hope to lay outa framework from which schools can ask good ques-tions and evaluate proposals by vendors, state agen-cies, district officials, and other interest groups.

A Matrix of LAN-WAN ConnectionsThe table on the next page is a framework that allowsus to usefully sort the cases that we found. "Type ofconnection" refers to whether the classroom PCs areterminals or are connected directly to a packet-switchednetwork. The three columns refer to the local in-frastructure. The lefthand column refers to stand-alone PCs, which, like the other columns, are con-nected to some remote services. The cases in the right

18 15

two columns have put their LANs to work in connect-ing to the remote services. The cases in the far rightcolumn, in addition, have a local server on the LANthat is used in communicating with people and ser-vices outside the school.

O

z

maintained by the Board of Education. The bulletinboard is in turn connected to the Internet through aleased line connection to NYSERNet, the New YorkState regional Internet network, but our interest hereis in how schools are making their connection.

Remote Services

Local Area Networks

Local Server

"Telecomputing' Networkmodem

"Portage"

Remotenetwork

Leased line tolocal campus

Local Internetserver

In this section, we describe cases of each model,beginning in the upper left of the matrix.

"Telecomputing"In the upper left cell of the matrix are all the schoolsinvolved in telecomputing or telecommunications,including schools with access CO the Internet, with theexception of the handful of schools that are of interestto this analysis. These schoolsuse a terminal type of connec-tion between their classroom(or other) PC and a host com-puter or remote service. In-cluded in this cell are theschools with LANs but withno connection between theLAN and the wide area net-work (e.g., where the com-puter with the modem is noton the LAN, or is on the LANbut no use is made of thatfact).

The diagram to the right

School

,,tes

ill

tr'

Network modemsNot all schools with terminal con-nections are underutilizing theirLANs. We found several schoolsusing a device called a networkmodem on the LAN. They still es-tablish a standard terminal connec-tion, but the network modem andspecial system software running onthe Macintosh computers let anyMac play the role of the terminal(one at a time) just as if the modemwere connected to it directly.

For example, teachers and stu-dents at Jefferson Junior High inOceanside, California, are activeusers of FrEdMail, CSUnet and

other education services. They have 30 Macintoshcomputers on an AppleTalk network in a lab with a fileserver (AppleShare).

The school connects to the local FrEdMail node,which now has a gateway connection to the Internetvia CERFnet, a regional network in southern Califor-nia. It also connects to CSUnet through a local statecollege campus. Both are terminal connections. Stu-

dents come into the lab and work ontheir word processing. When they are

7k)

Terminal

si

*MI5

Modem

(44410e

ready to send their message, they log ondirectly from their Mac. The availabil-ity ofa network modem lets many morestudents get involved in the projectsthan previously (in some projects thereis 100% involvement). The greater flex-ibility allows groups of students to ex-

NYC B. of Ed.

'41.4N.6:1PMMPPwwWW

shows a typical example. Aschool in New York City has a computer with amodem and calls into the bulletin board system (BBS)

Host a DSWits

Internet

16 19

Jefferson J.H.S.6114

terminal connection, can only support one terminal-host interaction at a time.

The "Portage"Back in the old days, fur traders would frequently haveto carry their canoes from one body of water toanother, a chore known as a portage. We identifiedtwo schools that carry on an analogous labor-intensive

process between the local andthe wide area network. Onecomputer is set up with themodem on it (or a networkmodem is installed on theLAN), and is used for commu-nication with outside electronic

mail services and databaseservices. The teacher logsonto the outside service us-

ing ordinary terminal emulation software, and copiesthe mail and data to the local disk. The teacher thenruns another program (e.g., a LAN mail system),retrieves the messages from the disk, and sends themover the LAN so that the internal community can haveaccess to them.

In the Computer Mini-school of the Ralph BuncheSchool in New York City, students are engaged in alarge number of projects, including collecting, shar-ing, and analyzing international data on changing sunshadow lengths; creating a guidebook which is sharedwith students in England; and obtaining and analyz-ing weather data from an information utility. Theteacher-in-charge has accounts on AppleLink,NYCENET (the New York City BSS), Dialcom, anda variety of other services and bulletin boards. Withinthe school, a local area network (EtherTalk andLocalTalk) connects some 60 computers (Apple II andMacintosh) and two AppleShare file servers.

FrEdMail Node

Internet

plore the network for appropriate projects, takingsome of that burden off the teacher.

The network modem scheme adds great flexibilityto the telecomputing paradigm because now anycomputer on the LAN can initiate communication. Incases where LAN computers are distributed aroundthe school, the network modem would allow all theclassrooms to share the very scarce resource of atelephone line. However, only one Mac at a time canbe a terminal since the modem connection, like any

Ralph Bunche School

GI/N

No-

Internet

20 17

A local mail system (Bank Street Writer withemail from Scholastic, originally designed by the-Earth Lab project) operates from the file server. Stu-dents and teachers can access their own "laboratory"on the file server and can communica, via the localemail system from any of the school computers. Forexample, students in the Shadows project send mes-sages to a "dummy" user called AGE (for Apple GlobalEducation network), sometimes mentioning a par-ticular addressee in the subject line. Once a day, theteacher checks the AGE account of the local mailsystem and transfers the messages to the appropriaterecipients on Apple Link.

From the perspective of the students and teachers,their email is connected to, the communities outsidethe school. Most are unaware of the behind-the-sceneswork of the teacher-in-charge who transfers the localmail to their destinations on the various networkservices. A notable advantage of the portage is theopportunity for the teacher-in-charge to monitor theoutgoing communication and to help direct incomingmessages (e.g., those addressed to the larger AGEcommunity) to appropriate student teams.

The previous three models were all variations ona terminal to host connection. We now move down tothe second row of the matrix and consider the modelsthat make use of network connections, beginning witha model involving a stand-alone PC.

Remote network connectionsA number of schools use SLIP software (described inan earlier section) to connect a single PC to their

School

ModemSerial Line IP

Internet regional. Once the connection is made, thePC is able directly to address any other computer onthe Internet; for example, to retrieve electronic mail,transfer files, or log in and run programs remotely.

Another (relatively easy to use) version of remotenetwork connections is AppleTalk Remote Access, aMacintosh System 7 extension that allows a remotecomputer to dial into an AppleTalk LAN and interactwith other computers on that LAN as though it wereconnected directly (with the limitation of the muchslower speed associated with modem connections overtelephone lines compared to that of a LAN). Forexample, teachers at the Stratton School in Arlington,Massachusetts call up using a Macintosh LC equippedwith a 9600 bps (bits per second) modem to a Macin-tosh on the LAN of a local corporation and are able tosend and retrieve electronic mail via the corporate mailserver. Since the connection is a network connection(transmitting AppleTalk packets across the phoneline), the school's LC is able to run a mail client thatmakes it possible to use the Macintosh interface for allreading and writing of mail. Since the corporate LANis connected to the Internet, the school LC (while thephone connection is open) is a full member of theInternet and can directly transfer files to or from anyotaer Internet machine. A remote network connec-tion can also be made from home to school. Forexample, the Ralph Bunche School teacher-in-chargeuses a network modem (Shiva NetModem) at theschool to call in from home to retrieve files or updaterecords on the school file servers.

Leased line to a local university campusWe found several examples of schools that were work-ing closely with a local university in a way that madethe school LAN an extension of the campus network.For example, McMillan Junior High School in Omaha,Nebraska has a largeApple-Talknetworkserving 80 to90 Macintosh stations and an AppleShare file server.The school uses a connection to a University ofNebraska at Omaha (UNO) computer for electronic

mail service and for access to files on otherInternet computers.13 Fourteen or 15 students

and about 20 staff (out of about80) have accounts on the UNOmachine. Macintosh users at theschool run a Telnet program to

establish virtual terminalconnections to the UNO

131=11)t$

a

Internet

18

May 1992 Tech. Rep.. No. 22 ,

machine. Once logged on, they can read and sendemail, and transfer files (using anonymous FTP) fromother Internet systems.

Students and teachers are involved in several for-eign language and social science projects. (One popu-lar use of the network is to access Supreme Courtdecisions from a University of Maryland machine.)Since it appears that the UNO host is oversubscribed,recent discussions at the school have explored thepossibilities of setting up a district server that wouldprovide file, mail, and Internet routing service, but nodecision has been made.

Davis H.S.

high school students are better able to individualizetheir research projects not having to stay lockstepwith the class.

Shortly, the services provided by this remote ma-chine (Internet mail and news) will be replaced by alocal server (running on a UNIX workstation) at theschool. UC Davis staff will manage this server re-motely until the school staff are ready to take on thisresponsibility themselves.

Local Internet serverSeveral schools have already taken the step thatMcMillan JHS and Davis HS are contemplating andhave set themselves up independent of local institu-tions. For example, Rocky Mountain High School inFt. Collins, Colorado has built a networked comput-ing environment that serves its district as well as theschool itself."

The school has a complex internal internetwork ofIBM PCs and Macintosh computers. The local net-work also includes two UNIX servers that provideemail and Internet news service for the school. Stu-

U.C. Davis

Another school set up as an extension of a localcampus network is Davis High School in Davis,California, which has a similar relationship to Univer-sity of California at Davis where about 150 of thestudents (out of 1200) and many of the staff haveemail accounts." In addition to email, the campusserver provides Internet news to the school. Projectsinvolve foreign language communication, research onabortion rights (including accessing remote data onSupreme Court decisions and congressional votingrecords), a science and math exchange with Russia,and remote supercomputing. The teachers find thatmany more students can be involved given this directconnection to UCD than with the single computerwith modem they used previously. Importantly, these

Internet

dents are engaged in a variety of projects, especially inthe area of envi. mment studies, using space imagery.Individual email and Internet access for the studentshas opened the way for individualized projects as wellas projects in which a collaborative group goes off toexplore a new Internet resource.

About two thirds of the current email accounts onthe UNIX servers belong to district users. The schoolexpects the number of mail accounts and remote usersto grow into the thousands as counselors, principals,other administrators, and other students find value inthese services. Meanwhile, local students and staff canaccess resources on the Internet directly from theirdesktop machines.

22 19

Rocky Mountain H.S.11.4

all the privileges of being a node on the network.The two cases are similar in that they require only

a phone line and an ordinary modem (2400 bps orhigher is recommended for the remote network). ThePC hardware requirements push toward the higherend for the remote network (e.g., the software is notavailable for an Apple II) than for the terminal-hostmodel (which requires only rudimentary terminalemulation software). The concern that many schools

have only Apple Ils is becoming less troubling giventhe current acquisition of higher-end machinessuch as the Macintosh. In any case, the Apple Its

can be networked locally with a singleMac acting as the network serverin a portage arrangement.

The remote network con-nection is certainly not as wide-

Internetrouter

Another example of a local Internet server is at theIllinois Math and Science Academy, a public, residen-tial school for talented students.16 Students maintaina Sun workstation and other computers that provideemail and other Internet services to the computer labsand resident halls. With full Internet access, studentsare involved in many projects and communicatingwith mentors at Argonne National Laboratory, FermiNational Accelerator Laboratory, and other scientificand research organizations. The IMSA campus net-work makes it possible for groups to collaborate onthese projects without face-to-face meetings.

Comparing the Six ModelsSetting up the models in a matrix makes it easier tocompare features and decide what the various technolo-gies are buying. As a way of understanding the value ofthe network connection, we can look at each columnof rr atrix in terms of the commonalties as well asthe differences between terminal and network con-nections.

The stand-alone classroom computer

In the first column, we have two models for connect-ing a single, non-LAN, PC to remote services. In the"telecom-puting" case, the PC becomes a terminal. Inthe remote network case, the telephone line becomesan extension of the packet network and the PC enjoys

Internetwot11111.

spread as the terminal connection, but there is reasonto believe that it will increase in use as more schoolsacquire access to the Internet. This type of "single userdial-up" service is offered by most, if not all, commer-cial Internet providers and regional networks today.

E

0a)z

Remote ServicesLocal Area Networks

Local Server

'Tof COm utl Networkmodem

"Portage"

fiernoksegtwork

Leased tine tolocal campus

Local Internetserver

LAN as access to the outsideThe middle column represents cases in which theLAN is used as an effective mechanism for expandingaccess to the wide area connection. With the LAN inplace, a student or teacher can sit down at any com-puter to call up a mail server or other remote resources,such as data on Supreme Court rulings. The immedi-ately obvious advantage of the network connectionover the terminal connection provided by the networkmodem is that many sessions can be carried on simul-taneously. The function of the router at the school is

20 23

to direct the appropriate packets off to the universitycampus network. The router does not care how manydifferent computers on the LAN are generating thepackets (i.e., up to the capacity of the phone line tocarry the traffic). The network modem, on the otherhand can only maintain a single terminal-host connec-tion at a time.

Remote Services

Local Area Networks

"TelocomputIng

Local Server!""'""Portage"

Remotenetwork

tease ltrio to Local Internetserver

A common limitation in these middle cases is thenumber of email accounts the school has available.This is not a technical limitation. It results from thecampus network administrator's not having the timeor resources to create hundreds or perhaps thousandsof email accounts for all the teachers and students inthe neighboring school. In all the cases we observedwhere a school depended on a neighboring universityor other organization, the helpful neighbor viewed thearrangement as temporary.'7 The plan was eventuallyto off-load the work of maintaining email accounts toa server administrator in the school or district. At least,

case, the direction ofin the network connectionmovement appears toward thematrix cell to the right.

We might characterizethese cases as using the LANto deliver the outside resourcesas illustrated in the diagram tothe right.

In these cases we observeda differentiation between thefunctions internal to the LANand the functions of externalaccess. In one case, for ex-ample, computers wererebooted in order to operateover the WAN, at which point

Supporting a local serverThe right- hand column represents the few schoolsthat maintain a server that supports communicationwithin the school as well as between the school and theworld outside.

00a)

00oa .A>.

Z

ca

1.---1;mote ServicesI Local Area Networks

Local Server

'TelocomputIng. Networkmodem

&tag°

Remotenetwork

Leased line tolocal campus

t,i5calInteinet

The portage model is a manual version of thefunctions served automatically by the Local Internetserver. In the latter case, a common addressing schemeis used for messages sent either locally or to distantdestinations.

The portage requires that the system administra-tor manually readdress messages going in either direc-tion across the external connection. Similarly, forremote database access, the administrator has to down-load the information of interest and place it on theschool server for local consumption. Besides avoidingthe inconvenient portage, the network case, of course,allows much greater immediacy of access to local andwide area resources from any computer. It also permits

activity that is not possible at all with aportage, such as remote login to distant

School

Ed

computers.The important similarity between

the two models is that local and distantcommunications are supported by the

same system from the students' andteachers' points of view (in the

portage, only the admin-

ii-± "1"-)0Agnasisiosy .0--11-1111040Nic

I University

.11:01:1

they could no longer access the local resources.

Internet

24 21

istrator sees two separate systems). The commitmentof effort to administer local mail and other networkservices makes it possible to give all students andteachers an email account.

In these cases, we find local communicationssupporting local project groups that are accessingremote resources. Teachers use the local communica-tions to coordinate their own work. The LAN is nowa medium for supporting communication and sharinginformation among the local community, not just ameans of access to the wide area resources.

The contrast between the middle column casesand the right column cases, which we attempt tocapture in the two diagrams illustrating the flow ofcommunication, is reminiscent of the difference wenoted earlier between the instructional delivery andcommunicative uses of school LANs. The local serverallows the school to go beyond using the LAN simplyto provide access to the outside. The LAN can nowserve an organizational function in the school,mediating the communication among students andteachers.

22 25

. cc . ep.

Prospects for the Future

How do we move from a situation in which only a minority of schools have even a terminalconnection for a single classroom PC to a situation in which there is universal access to anational data network such as the NREIV? We believe the prospects for universal access arevery good as long as schools can make use ofcurrent and planned investments in local, district,and state education networks. We see the terminal-host paradigm as essentially a dead endIfthe goal is to make resources available to all teachers and students, the centralized computingparadigm (the basis for the terminal-host connections) cannot handle the task. In thisconcluding section, we suggest the next steps that must be taken toward a school internetwork.

Build Remote Network ConnectionsA number of paths might be recommended to getfrom the upper left cell of the matrix to the lower rightcell. One path is the "high road" along the top row ofthe matrix. This path continues the terminal-hostparadigm. When LANs are acquired, network mo-dems are used to distribute access to the phone line andwhen local servers are used, material is portaged.

Remote ServicesLocal Area Networks

Local Server

"Taux.:Imputing" Networkmodem

"Portt!g0"

Remotenetwork

Leased line tolocal campus

LocatinternetserviV

This route has major limitations because the re-mote and local systems remain essentially indepen-dent, requiring the manual portage of the materialfrom one to the other. The transition from the portageto the Internet server requires considerable retooling,since PC software in use locally may not function onthe wide area network.

Another path, shown above, is the "low road"along the bottom row of the matrix. In this case, wewould begin by migrating from the upper left to thelower left; that is, not assume the initial availability ofLANs, but begin developing the expertise and thesoftware to support true network connections in schoolswith stand-alone computers.

This direction may be more workable. With theinstallation of a LAN and router, the same software,resources, and style of interaction that is familiar from

Remote Services

Local Area NetworksLocal Server

"Telecomputing' Networkmodem

"Portage"

..netw

4Ased.:0110.1. P0 tfariet

the single connected computer is simply extended tothe other computers. An email client program, forexample, that is used on the one networked machinecan now be used on any of the machines on the LAN.With the later introduction of a local server, the samenetwork interactions are simply brought closer andoffered at a faster speed over the LAN.

This transition path argues for the aggressivemigration from terminal connections to remote net-work connections in education. Some of the softwarefor these connections, which are currently offered as acommercially available service, may require im-provement for use by teachers who are new at net-working. The migration path is worth developing,however, since it will open up to schools the kind ofconnection that can support standard forms of clientsoftware.

Address Security and Management ConcernsThe open access that comes with internetworkingraises concerns about how to keep administrativerecords secure and how to monitor student communi-cation. Much of the existing state and district levelnetwork infrastructure (as well as many of the schoolLANs) supports administrative communication anddatabases of student records. The resistance that we

26 23

have found in discussions of using existing adminis-trative networks for instruction often revolves aroundfears of students breaking in and compromising thoserecords. Instructional networks in the schools couldalso be prey to "hackers" intent on destroying orstealing data or planting viruses on school computers.Certainly such concerns have been addressed in mili-tary and business networks and by the Internet com-munity (e.g, Holbrook tic Reynolds, 1991), but schooltelecomputing has not made use of the level of tech-nology available in other sectors. The design of anational educational network will have to incorporateadequate security mechanisms.

Educators also expressed concerns about manag-ing communication from the school to the outsideworld or even communication within the school.Occasionally, the worry is that students or teacherswill make use of the system for communication unre-lated to instruction. For example, students may dis-cuss socially sensitive issues or teachers may conductunion-related business. In business, research, and aca-demic settings, extracurricular communication is sel-dom controlled. Provocative communication via emailis dealt with as it would if expressed in other media. Itis not clear that schools should be any different in thisrespect. There is a difference, however, in the case ofcommunication with the outside world. Even teacherswho are deeply committed to empowering students'self-expression question the wisdom of allowing un-regulated student communication on national andinternational networks in terms ofboth the reputationof the school and the needs of the students for helpwith clarifying their communication. An advantage ofthe telecomputing model in which only the teacherhas an account on the outside service is that allmessages that cross the wire can be monitored. Theteacher can work with children, who cannot be ex-pected to know the rules ofnetwork discourse, to cleanup their communication and appropriately distributethe incoming messages. While many teachers will feelcomfortable giving many students unrestricted access,a communications system for schools, to be acceptableto the majority of teachers and administrators, willhave to provide for the possibility of monitoringstudent messages.

Develop an Integrated ServerCurrently, there is no server on the market, such as theone illustrated in the hypothetical scenario in the

previous section, appropriate for helping schools andschool disticts connect to the Internet. A school servershould integrate several commonly needed servicesand provide an adequate system for administering ormanaging these services. Such a server might be imple-mented on a relatively inexpensive computer, such asa 486 PC running a version of the UNIX operatingsystem. Besides database, file, news, and mail services,the server should also provide a standard Internetrouting service that allows anybody at a PC on theLAN to communicate over the entire Internet.

Most of the network services software that wouldbe needed is readily available for integration on such aplatform. A management system, based on a sophisti-cated database tool, could integrate the administra-tion of these services and provide the kind of securityand management of access privileges required in theschool setting. The development of the managementsystem in a way that takes advantage of existingadministrative databases for the creation and updatingof teacher, student, and other school accounts, thatcan be easily maintained as a part-time activity in theschool, and that allows appropriate distribution ofthese tasks among coordinators and classroom teach-ers is perhaps the most critical component of a schoolserver.

Develop Client SoftwareAn important reason for moving quickly toward truenetwork connections is that it will open the way foreducational software developers to create networksoftware that will be usable regardless of where in theinternetwork the classroom PC finds itself. To allowthe smooth scaling up from a single stand-alone com-puter with a modem in a school to the districtwideInternet, the fundamental relationship between theclient software running on the classroom computerand the software running on the server must be thesame regardless of whether the server is located in thecomputer lab next door or available by dialing up tothe regional network center in another state.

There are near-term prospects for true networkedsystems coming into the schools as the commercialstandards for client-server software interfaces are de-veloped and adopted by manufacturers. Such "net-work-enabled" software, whether a word processor, adatabase interface, or a simulation, is able to com-municate with resources anywhere on the Internet. Inthis way, a spreadsheet or a simulation can directly

24 27

send or receive electronic mail, or a HyperCard stackcan obtain and display data from a remote computer.A major advantage of network connections is the fargreater ease of interaction with network services theyafford. Using standard internetwork protocols makesit possible to have a completely open and generalnetwork-enabling strategy in which all educationalsoftware developers can participate.

The network paradigm will not be a lowest com-mon denominator solution to school networking. Wedo not believe that there is a direct trade -off betweencost of the technology and our ability to provideuniversal access. The lowest cost, short-term solution,a simple terminal-host system using very low speeddial-up lines, will not be able to take advantage ofexisting and planned investments in district-wide net-works to lower costs, and will not be able to encompassmultiple applications to raise the value of the invest-ment. It is also far less likely that easy-to-use diem

software will develop for the terminal-host paradigmin the way it is all dy developing for networks.

The history of school LANs and WANs is differ-ent from that in other sectors. It is not simply thatschools lag behind. The functions of the technologyhave been different, resulting in two distinct tracks; inother domains, LANs and WANs evolved together.Schools remain different functionally, organizationally,and in scale from business, research, military, andother environments in which networks have beendeployed. Current solutions in these sectors will notnecessarily be adequate for schools but schools canlearn from them and borrow technology as needed. Itmay well be the case that future network solutions forschools will be borrowed by industry as we confrontfor the first time the issues of scale and ease of use andbegin creating organizations in which people can learnthrough communication.

Notes

1. The National Research and Education Network(NREN) is a component of the High Performance Com-puting Act.

2. CTP mailed a 6-page questionnaire to a randomsample of 1,000 school sites (of the over 9,000 school sitesin California) in April 1989.

3. One case that approached such a use had a smallsingle-purpose LAN that connected several terminals CO adatabase in the state capitol. But this set-up was indepen-dent of the main LAN in the computer lab.

4. Packet switching was subsequently developed formany transmission types, including radio, satellite, andlocal area network technologies such as Ethernet.

5. The first packet switches that formed the ARPANETbackbone connected computers at UCLA, the StanfordResearch Institute, UC Santa Barbara, and the Universityof Utah. Before the military sites split off to form a separatenetwork in 1983, over 100 ARPANET packet switchesconnected hundreds of computers at university campuses,research labs such as the Rand Corporation and LincolnLabs, other companies engaged in research and develop-ment such as Xerox, DEC, and Bolt Beranek and Newman,and government sites, all of which had U.S. Government-sponsored projects.

6. The Internet technology, commonly referred to asTCP/IP, includes protocols that specify how computersand their applications communicate (notably, Transmis-sion Control Protocol or TCP) and protocols that specifyhow data traffic can be routed over interconnected net-works (notably Internet Protocol or IP).

7. K 12net discussion groups can also be accessed asnewsgroups.

8. For example, machines with large shared resources,such as disk space or databases, centralized management,and so on.

9...A more recent standard being implemented is thePoint-to-Point Protocol (PPP).

10. This is the case, even when a public or commercialdata network serves as an intermediary between the termi-nal and host. For example, when a teacher calls CompuServeor any of the other national "information utilities," he dialsa local number of a specialized computer, which takes theincoming information. The information is turned intopackets and sent out over a national network (usually anX.25 packet-switched network) to the host computer some-where in the country. At the far end, the packets aredisassembled into a stream of bits so that, from the point of

28 25

view of the classroom computer and the host, there is stilljust a terminal-host connection.

11. An important difference between a dial-up line anda leased line is the latency in response for highly interactiveapplications, since a dial-up connection would have to beestablished before communication takes place. Leased lineswill be advantageous for synchronous interaction, remotecomputing, or where finer grained interacdons vvith Inter netservices are desired. A leased digital line allows for muchhigher speeds of data transmission but may not be requiredimmediately by schools. In a large state, multiple Internetconnections will be desirable as traffic increases.

12. These include: Jostens Learning Corp.; ComputerCurriculum Corp.; Computer Networkin g S pecialists, Inc.;Wicat; Novell; IBM; Velan; Wasatch Education Systems;and Ideal Learning.

13. A gateway (Cayman's GatorBox) connects theLocalTalk nerwork (the physical network supporting theMacintoshes) to the Ethernet interface ofan internet router.The McMillan router is then linked to a router at UNOusing a leased 56 Kbps line.

14. Davis High School students and staff useMacintoshes on a LocalTalk network, but run TCP/IP ontheir Macs and are connected to the Internet. One of theMacintoshes acts as a file server; it runs AppleShare, andstudents run Mac-based email (MacMH now, Eudorasoon), terminal emulation (Telnet), and file transfer (FTP)applications from the server. In this way, incoming mailand files are stored directly on the server in one step. Agateway (Shiva FastPath IV) connects the LocalTalk net-work to the Ethernet interface of an Internet router (Proteon)in the school. The Davis High router is then linked to arouter at the University of California (UC Davis) using aleased 56 Kbps line. A network services VAX host on theUC Davis campus network provides accounts for theexchange of mail between Davis High and the Internet.These are not accounts that allow people to log in; they onlyallow email exchange by the PC exchange protocol (POP).

15. Within the school, students and staff useMacintoshes and an AppleShare file server on a LocalTalknetwork, which is gatewayed to the school Ethernet (Cay-man GatorBox). Users of the school's business lab work onPCs served by a Novell NetWare server, all of which are

directly on the Ethernet. All networked Macintoshes andPCs have Internet terminal emulation (Telnet) capability.Two IBM RS/6000 ADC (UNIX) machines on the Ethernetprovide email, selective Internet news feeds, and Telnetservices to the school and to elementary, junior high, andadministrative users in the larger district. This is madepossible by two wide area connections. First, an Internetrouter (Cisco) on the Ethernet connects the school networkvia a 56 Kbps dedicated line to a regional Internet router atColorado State University. Second, a dedicated line andsome protocol converters allow terminal users of the district'sVAX cluster (which does not support TCP/IP) to access theschool's UNIX servers.

16. IMSA supports a heterogeneous local area networkenvironment. Their network is connected to an Internetrouter at Argonne National Labs (an NSFNET high speedbackbone node site) by means of a local Internet router(Cisco) and a leased 56 Kbps line. The IMSA networkenvironment is built around two Ethernets (one educa-tional, one administrative connected to each other by theirrouter) and an extended LocalTalk network connected tothe Ethernets by gateways. The Academy boasts a numberof computing resources, several ofwhich are used as servers.For example, there are two student and two administrativeNetWare file servers. Servers are designated as either edu-cational or administrative; administrators can access bothtypes, but students and the rest of the Internet can onlyaccess the educational servers. IMSA students and staff useMacintoshes and PCs on the network, as well as a handfulof UNIX workstations of various types; all support TCPand can reach any Internet host directly when necessary. AUNIX (Sun) server supports most of IMSA's electronicmail locally and all email to and from the Internet. It alsoprovides a local bulletin board service ("notesfiles"). Avariety of client mail programs, which communicate withthe UNIX server, are used on the Macintoshes and PCs(QuickMail, Eudora, Pegasus); some users use terminalemulation (Telnet) to access their email directly on the mailserver.

17. This view was echoed by many participants at arecent Boston workshop, "NEARnet and Grades K-12:Educational Innovation through Partnerships."

26 29

Toy es .-'ep. lo.

References

California Technolo Project Assessment Team (1990).An assessment of educational technology applications inCalifornia public schools. Chico, CA: California StateUniversity.

Cerf, V. (1989). Requiem for the ARPANET. ConneXions,3(10).

Cole, M., & Griffin, P. (1987). Contextual factors ineducation: Improvingscience and mathematics educationfor minorities and women. Madison, WI: WisconsinCenter for Educational Research.

Corner, D. (1990). Internetworking with TCP/IP.Englewood Cliffs, NJ: Prentice-Hall.

Dern, D. P. (1989). The ARPANET is twenty: What wehave learned and the fun we had. ConneXions, 3(10), 2-10.

Dern, D. P. (1992, February). Applying the Internet.BYTE, 111-118.

Electronic Learning (1992, March). The ILS Mac Pac.Electronic Learning (Special Edition).

Goldman, S. V., & Newman, D. (in press). Electronicinteractions: How students and teachers organizeschooling over the wires. Interactive Learning Environ-ments.

Guruge, A. (1992, February). SNA and LAN internets: Along walk to the altar. Network World; 9(6).

Holbrook, P., & Reynolds J. (1991). Site Security Hand-book. Internet Request for Comments 1244, FYI 8.

Hunter, B. (1992a). Linking for learning: Computer-and-communications network support for nationwide in-novation in education. Journal ofScience Education andTechnology, I.

Hunter, B. (1992b. April). Internetworking: The coordinat-ing technology for systemic change in education. Pre-sented to Subcommittee on Select Education, Com-mittee on Education and Labor, U.S. House of Repre-sentatives, Congressional Staff Forum on "Designingan educational technology infrastructure: Defininthe federal role in the dissemination of information.

Kehoe, B. P. (1992). Zen and the art of the Internet: Abeginner's guide to the Internet.

Kurshan, B. (1990a). Home market for educational onlineservices(Research Report). Educorp Consultants, 4940Buckhorn Road, Roanoke, VA.

Kurshan, B. (1990b). Statewide telecommunications net-works: An overview of the current state and the growthpotential (Research Report). Educorp Consultants,4940 Buckhorn Road, Roanoke, VA.

Kurshan, B., & Harrington, M. (1991). Statewide educa-tion networks: Survey results. Educorp Consultants,4940 Buckhorn Road, Roanoke, VA.

Kurshan, B. (1992). The global classroom-Changing theway we learn and teach. Educorp Consultants, 4940

Bucichorn Road, Roanoke, VA.LaQuey, T. L. (1990). The user's directory of computer

networks. Bedford, MA: Digital Press.McAnge, T., Harrington, M., & Pierson, M. E. (1990). A

survey of educational computer networks. Blacksburg,VA: Virginia Polytechnic. Institute and State Univer-sity.

Mageau, T. (1990). ILS: Its new role in schools. ElectronicLearning, 10(1), 22-32.

Martin, J. (1991). There's gold in them char networks!Internet Request for Comments 1290, FYI 10.

Melmed, A., & Fisher, F. D. (1991). Towards a nationalinformation infrastructure: Implications for selected so-cial sectors and education. New York: Center for Edu-cational Technology and Economic Productivity, NewYork University.

Newman, D. (1990). Opportunities for research on theorganizational impact of school computers. Educa-tional Researcher, 19(3), 8-13.

Newman, D. (in press). Technology as support for schoolstructure and school restructuring. Phi Delta Kappan.

Newman, D., Goldman, S.V., Brienne, D., Jackson, I., &Magzamen, S. (1989). Peer collaboration in com-puter-mediated science investigations. Journal ofEdu-cational Computing Research, 5(2), 151-166.

Newman, D., & Reese, P. (1990). Using a local areanetwork for sharing data. In EPIE Institute (Eds.),Report on integrated instructional systems. Water Mill,NY: EPIE Institute.

NYSERNet, Inc. (1991). New user's guide to useful andunique resources on the internet (Version 2.0). Syracuse,NY: Author.

Office of Science and Technology Policy. (1991). Grandchallenges: High performance computing and communi-cations. A report by the Committee on Physical, Math-ematical, and Engineering Sciences to supplement thePresident's FY 1992 Budget.

Reilly, R. (1992). K12net: International educational net-work.

Roberts, N., Blakeslee, G., Brown, M., & Lenk, C. (1990).Integrating telecommunications into education.Englewood Cliffs, NJ: Prentice Hall.

Sherry, M. (Ed.). (1990). The integrated instructional sys-tems report. Water Mill, NY: EPIE Institute.

Solomon, G. (1992, March). The most complete guideever to telecommunications. Electronic Learning, 18-28.

St. George, A. (1992). A voice for K-12 networking.Research and Education Networking, 3(1), 10-12.

Wood, R, & Mensch, H. (1991). The Athena networkservices environment: An overview. Presented atUniforum.

30 BEST COPY AVAILABLE27